Have an AI proof of concept built
You do not want research into what AI could do, but something that works with your own data. Within weeks, not quarters — so you decide based on a running thing whether to continue.
What a proof of concept actually is
A proof of concept is the smallest version of an idea that proves it works — with your data, in your processes, in the hands of your people. Not a demo on made-up examples, and not a pilot that has to be production-ready from day one. It answers one question: does this deliver enough to keep building?
That distinction matters more with AI than with ordinary software. A language model nearly always looks good in a demo. The question is whether it holds up on your own documents, with your exceptions, used by someone who was not waiting for it. You only find that out once it runs.
So I would rather build for a week than research for a month. A report on what AI could do takes longer, convinces fewer people and ages faster than something you can click open.
How it works
01
Choosing what is worth it
We collect ideas from across the organisation and pick the most promising together. Not the one with the prettiest demo, but the one most likely to be used.
02
Building with your data
I build it into something that runs in your context. So you learn not what AI can do in theory, but what it delivers for you.
03
A clear decision
Every proof of concept ends in keep building, adjust, or stop — with a working thing as the basis for that choice.
What that looks like in practice
Three things that started out this way. They all run — click through and see for yourself.
B2B visitor identification
Revealu
98% of companies leave your site without filling in a form. Revealu matches traffic to company data and shows in real time who is browsing, which pages they read and how warm they are — no cookies, no form. A full alternative to Leadinfo.
revealu.nlAdoption software for animal shelters
Animalz
One popular animal draws thirty replies in a few hours, all answered by hand. Animalz sends a personal message on every status change and pre-sorts replies, so the hours go to the animals instead of the inbox. Open source, free at its core.
animalz-iota.vercel.appDictation in your own writing style
TokTalk
Speak instead of type, with text that comes out sounding like you write. Runs on European AI with EU data storage; here privacy is not a footnote but the product. Mac and web, paid from €4.50 a month.
toktalk.coWhen a proof of concept is not the right step
Better said upfront than invoiced afterwards.
- The specification is settled and the problem is proven. Then you need building capacity, not validation — hiring a developer is faster and cheaper.
- It is mostly maintenance on an existing system. I build new things; maintenance is not where I make the difference.
- Nobody in the organisation may know about it yet. Without people using it you prove nothing — that makes it a demo, not a proof of concept.
a poc is not a promise, it is proof →
What it costs
I charge € 150 per hour on 8-hour days. Two days a week for three months comes to € 28,800 excluding VAT. On the pricing page you can work out what your engagement costs.
Work out what your engagement costsA working proof of concept — or that phase costs you nothing
And you can stop at any decision point — no obligations
What people ask about this
- How long does a proof of concept take?
- A working first version is usually there within a week. The more complex the problem, the longer it takes — but the premise is that you see something run within weeks instead of reading a report. The weeks after that are about sharpening it based on what your people think of it.
- Who owns the code that comes out of it?
- You do. I do not believe in IP on code: it is increasingly a commodity. What you keep is not the code, but the insight into whether this was worth building.
- Do we work with our real data?
- Preferably yes, because that is the only way to know what it delivers for you. Where the proof of concept and the data live is something we decide together based on your policy; I prefer an isolated environment of my own, so we do not get stuck on internal policy. A data processing agreement is part of onboarding.
- And afterwards — do you finish it?
- You decide who picks it up. I can recommend good developers, or help your own team or agency extend it. I stay available if that is needed or wanted, but what I am good at is getting things moving — so usually everyone is better off if I keep that role.
in closing
Want to see if this fits?
One conversation to find out whether there is something worth building — and if there is not, I will say so.
€ 150 per hour · 1 to 3 days a week · you can stop at any time